A chip testing and analysis method
By identifying the functional area and degradation coefficient of the chip, monitoring the dynamic processing volume and performance in real time, and dynamically adjusting the testing strategy, the problems of low chip testing efficiency and mis-testing in the existing technology are solved, and efficient and accurate chip testing is achieved.
Patent Information
- Application Number
- CN202411119245.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-15
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-08-15
AI Technical Summary
The existing chip testing technology cannot be adjusted according to the performance of the chip, resulting in low error testing and testing efficiency, and the inability to detect chips with abnormal tests in a timely manner.
By determining the chip's functional area, actual processing volume, processing interface information and degradation coefficient, identifying the problem area, and monitoring the dynamic processing volume and performance degradation in real time, dynamically adjusting the test strategy to ensure the accuracy and efficiency of the test.
It realizes accurate evaluation of chip performance and timely discovers performance degradation, optimizes testing processes, improves testing efficiency and accuracy, and ensures the stability and reliability of chip performance and product quality.
Smart Images

Figure CN118641935B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of chip testing, and in particular to a chip testing and analysis method. Background Art
[0002] Scan test based on automatic test equipment (ATE) is an essential step before the chip leaves the factory. Scan test is an efficient structured test method, which is of vital importance to ensure the quality of the chip. At the same time, scan test occupies a lot of machine resources and test time, which is often the bottleneck of the test and has a high impact on the test cost.
[0003] For example, Chinese patent publication number CN116774018A discloses a chip testing method, device and electronic equipment, which relates to the field of chip testing technology, including: determining multiple functional areas in a packaged chip to be tested; determining an allocation scheme of input and output ports of the chip to be tested according to a preset allocation rule; wherein the allocation scheme includes a correspondence between input and output ports and functional areas; the allocation rule includes: at least one or a group of input and output ports are allocated to multiple functional areas for sharing; based on the allocation scheme, one or a group of input and output ports are controlled by a port multiplexing module to selectively connect to one of the multiple functional areas; a test vector of an automatic test device is input into the connected functional area through one or a group of input and output ports, a scan link in the connected functional area is tested, and a test result is output through one or a group of input and output ports; however, the prior art cannot make adjustments according to the performance of the chip itself when testing the chip, resulting in easy occurrence of mistests when testing the chip.
[0004] For example, Chinese patent publication number CN115629296B discloses a chip testing method, device, equipment and storage medium, wherein the chip testing method includes: performing a capacity test on the chip and obtaining the capacity test result of the chip; performing a first test on the chip that passes the capacity test and obtaining the first test result, wherein the first test result includes the status information of the storage block; for the chip that fails the first test, determining the abnormal storage block based on the status information of the storage block; for the chip that fails the first test, shutting down at least a part of the abnormal storage blocks in the chip, and performing a second test to obtain the result of the second test. However, the prior art does not improve the efficiency of testing the chip by only simple abnormal identification when testing the chip, resulting in the failure to timely discover the chip with test abnormality. Summary of the invention
[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: a chip testing and analysis method, comprising: S1, determining multiple functional areas of the chip to be tested after packaging, and testing the input and output ports of the chip to be tested according to a preset allocation rule.
[0006] S2, determining the actual processing capacity, processing interface information and adopted test parameters corresponding to the chip when testing the input and output ports.
[0007] S3, determining the degradation coefficient of the chip under different workloads according to the actual processing amount and processing interface information of the chip.
[0008] S4, setting the scanning time of the chip, and recording the number of channels that the chip can connect to within the scanning time as the channel connection quantity; determining the test signal and test protocol corresponding to each chip.
[0009] S5, identifying the problem areas inside the chip based on the length of time the chip spends on processing tasks, the degradation coefficient, and the amount of channel connections.
[0010] S6, monitors the dynamic processing volume and performance degradation of the ports in the problem area in real time, and adjusts the chip testing strategy.
[0011] The beneficial effects of the present invention are as follows: First, the present invention determines multiple functional areas of the chip to be tested after packaging, and tests the input and output ports according to preset allocation rules; then, by real-time monitoring of the actual processing volume, processing interface information and test parameters of the chip, the degradation coefficient of the chip under different workloads is calculated; then, according to the chip's scan time, channel connection volume, test signal and test protocol information, the problem area inside the chip is identified; finally, by real-time monitoring of the dynamic processing volume and performance degradation of the port in the problem area, and based on this information, the test strategy is dynamically adjusted to ensure the accuracy and effectiveness of the test.
[0012] 2. The present invention can comprehensively evaluate the performance of the chip under different working conditions by real-time monitoring of the dynamic processing volume and performance degradation of the chip.
[0013] 3. The present invention can timely detect chip performance degradation through real-time monitoring and dynamic adjustment of test strategies, and take corresponding measures to repair and optimize.
[0014] 4. The present invention can optimize the test process according to the actual changes in chip performance by dynamically adjusting the test strategy, thereby improving test efficiency and accuracy.
[0015] 5. The present invention can ensure the stable and reliable performance of the chip and improve the quality and reliability of the product by timely discovering and solving the problem. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] The present invention is further described below in conjunction with the accompanying drawings and embodiments.
[0017] Figure 1 The present invention is a flowchart of a chip testing and analysis method.
[0018] Figure 2 It is a flowchart diagram of steps S21-S23 of a chip testing and analysis method.
[0019] Figure 3 It is a flowchart diagram of steps S51-S54 of a chip testing and analysis method.
[0020] Figure 4 It is a flowchart diagram of steps S61-S63 of a chip testing and analysis method. DETAILED DESCRIPTION
[0021] The embodiments of the present invention are described in detail below. The embodiments described below are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. If no specific techniques or conditions are specified in the embodiments, the techniques or conditions described in the literature in the art or the product specifications are used.
[0022] See also Figure 1 , a chip test analysis method, comprising: S1, determining multiple functional areas of a chip to be tested after packaging, and testing the input and output ports of the chip to be tested according to a preset allocation rule.
[0023] S2, determining the actual processing capacity, processing interface information and adopted test parameters corresponding to the chip when testing the input and output ports.
[0024] S3, determining the degradation coefficient of the chip under different workloads according to the actual processing amount and processing interface information of the chip.
[0025] S4, setting the scanning time of the chip, and recording the number of channels that the chip can connect to within the scanning time as the channel connection quantity; determining the test signal and test protocol corresponding to each chip.
[0026] S5, identifying the problem areas inside the chip based on the length of time the chip spends on processing tasks, the degradation coefficient, and the amount of channel connections.
[0027] S6, monitors the dynamic processing volume and performance degradation of the ports in the problem area in real time, and adjusts the chip testing strategy.
[0028] First, multiple functional areas of the chip under test after packaging are determined, such as clock circuits, memory, logic circuits, etc. Then, according to the preset allocation rules, the input and output ports of the chip under test are tested. This includes checking the connectivity of the ports, signal transmission quality, etc.
[0029] During the input and output port testing process, the actual processing volume of the chip, processing interface information, and the test parameters used are recorded. This information is crucial for subsequent performance analysis and test strategy adjustment.
[0030] According to the actual processing volume and processing interface information of the chip, combined with the workload of the chip, the degradation coefficient of the chip under different workloads is calculated. The degradation coefficient reflects the trend of chip performance changes over time and workload.
[0031] Set the scanning time of the chip, and record the number of channels that the chip can connect to within the scanning time as the channel connection quantity. At the same time, determine the test signal and test protocol corresponding to each chip. The test signal and protocol should be able to fully cover the functional area of the chip to ensure the accuracy and completeness of the test.
[0032] Based on information such as the length of time the chip spends on processing tasks, the degradation coefficient, and the amount of channel connections, problem areas within the chip can be identified. These problem areas may be the key to performance degradation and require special attention and adjustment.
[0033] For the identified problem areas, the dynamic processing volume and performance degradation of the ports are monitored in real time. This can be achieved through dedicated monitoring tools or systems to ensure the real-time and accuracy of the data.
[0034] like Figure 2 As shown, in step S2, when obtaining the actual processing capacity and processing interface information of the chip, the processing method for the chip includes: S21, setting the test parameters of the input and output ports according to the design specifications and application scenarios of the chip, including data transmission rate, interface type, and signal level.
[0035] S22, sending a preset signal sequence to the input port of the chip through the test software.
[0036] S23, collecting and recording the actual processing volume of the chip in real time; comparing the collected actual processing volume with the preset processing volume of the chip to verify whether the actual processing volume corresponding to the corresponding processing interface information meets expectations.
[0037] The actual processing capacity mentioned above usually refers to the amount of data or tasks that the chip can process under specific conditions. This may include the number of operations that can be performed per second (such as the number of addition, multiplication and other operations), the data throughput that can be processed (for example, the number of bytes that can be transmitted or processed per second), or the number of tasks that can be processed in parallel.
[0038] This indicator helps to understand the performance of the chip under different workloads and is an important basis for evaluating the chip's computing power, processing speed and efficiency.
[0039] Processing interface information involves the way and specifications of the chip to communicate with external devices or other chips. This includes the type of interface (such as SPI, I2C, UART, PCI-E, USB, etc.), data transfer rate, supported protocol standards, physical characteristics of the interface (such as the number of pins, voltage levels, etc.), and any specific communication timing requirements.
[0040] Knowing this information is critical to ensuring that the chip can be correctly and efficiently integrated with other system components, and it also affects how the test equipment is set up and the test scripts are written during the test process.
[0041] The processing method for verifying whether the actual processing volume corresponding to the corresponding processing interface information meets the expected requirements in step S23 includes: sending a preset signal sequence based on the input port of the chip, obtaining a first test result about the signal sequence, the first test result including a basic processing volume that the chip can process under the current workload, and determining the ratio of the basic processing volume to the actual processing volume of the first test result based on the type and protocol of the corresponding processing interface information; when the ratio of the basic processing volume to the actual processing volume of the first test result is less than the preset processing ratio, determining the difference between the actual processing volume and the preset processing volume under different workloads, and generating a difference change curve; determining the expected effect corresponding to the actual processing volume based on the change trend of the difference change curve over time; if the ratio of the basic processing volume to the actual processing volume of the first test result is greater than the preset processing ratio, identifying the corresponding processing interface information and the corresponding input and output ports, and using the corresponding processing interface information and the input and output ports as outputs.
[0042] When verifying the expected effect corresponding to the actual processing volume, the maximum and minimum points in the difference change curve are determined based on the obtained difference change curve, and the fluctuation range between the maximum and limit points within the corresponding time is output as the current expected effect, and the corresponding maximum and minimum points are marked.
[0043] The marked maximum and minimum points can assist in the subsequent processing of the actual processing volume, and can quickly identify the degree of influence of the actual processing volume under the influence of parameters such as time and degradation coefficient.
[0044] Under normal circumstances, the module determines the degradation coefficient by calculating the ratio of the actual processing volume to the benchmark processing volume. However, when different workloads appear, and when the chip test environment and temperature change, the working efficiency of the corresponding chip will be affected, and this impact may even cause a certain burden on the power supply voltage. As a result, when identifying the workload, it is necessary not only to process the current data according to the chip's usage time and actual processing volume, but also to consider the aging speed of the chip itself, and the time the chip's performance can be maintained and the time required for abnormalities to occur under degradation tests such as pressure and temperature that the chip can accept, as well as to determine the chip's working status and deduce the impact of the corresponding degradation coefficient based on this part of the working status.
[0045] In order to comprehensively evaluate the change in chip performance, the degradation coefficient obtained in step S3 includes a performance degradation coefficient and a thermal sensitivity performance coefficient.
[0046] The performance degradation coefficient is obtained by obtaining the degraded current, power, and delay under the corresponding workload, and integrating the index values of the degraded current, power, and delay into a comprehensive degradation coefficient as the performance degradation coefficient.
[0047] The performance degradation coefficient is expressed as: ;in, is the performance degradation coefficient, is the initial power, which is used to indicate the power value of the device when it is working normally. is the degraded power, is the power adjustment factor, is the initial current, indicating the current value for normal operation, is the degraded current, is the regulation factor of the current, is the initial delay, which indicates the delay of normal operation of the device. is the degraded delay, is the adjustment factor for the delay.
[0048] In order to improve the recognition effect of the degradation coefficient, the influence of the temperature on the current chip is determined according to the temperature change.
[0049] Thermal sensitivity performance coefficient, determine the thermal sensitivity performance coefficient based on the temperature obtained under the corresponding workload; determine the degradation rate corresponding to the thermal sensitivity performance coefficient based on the rate of change of the thermal sensitivity performance coefficient; the thermal sensitivity performance coefficient is mainly used to determine at what temperature the current device will experience performance degradation. By identifying this temperature, the chip can be assisted in degradation testing.
[0050] ;in, is the thermal sensitivity coefficient, is the initial operating temperature, is the ambient temperature, is the rate of change of temperature with time.
[0051] When testing the difference between the initial operating temperature and the ambient temperature, the initial operating temperature value obtained at this time is not equal to the ambient temperature value by default.
[0052] The degradation rate for acquisition is expressed as: ;in, is the rate of change of thermal sensitivity coefficient with temperature, is the degradation rate, is the frequency factor, is the activation energy, is the gas constant, is the absolute temperature.
[0053] The frequency factor and activation energy obtained above are used to describe the rate factor and the required energy threshold of the physical or chemical change process inside the chip; they are indirectly derived or fitted through specific test procedures, such as temperature cycle test, voltage stress test, etc.; these parameters can be used to evaluate the stability and reliability of the chip under different working conditions.
[0054] The gas constant represents the thermal conduction or thermal convection constant of the simulated chip during chip testing. These constants are usually determined based on material properties and environmental conditions, and may be obtained through thermal simulation or experimental measurement. They are used to indicate the phenomenon of heat conduction in the gas around the chip due to the increase in chip temperature during the corresponding test. This helps understand the behavior of gas in the heat transfer process, thereby more accurately simulating the thermal environment of the chip at work; and determines the test conditions of the chip in a specific gas environment to ensure the consistency and repeatability of the test.
[0055] Absolute temperature refers to the operating temperature range required by the current chip when it is working normally. For example, some chips may need to work normally in an environment of -40°C to +85°C. The absolute temperature at this time refers to this range of values. At the same time, the absolute temperature will be determined based on a comprehensive consideration of multiple factors such as the chip's material properties, circuit design, and heat dissipation capabilities to determine the operating temperature range of the corresponding chip.
[0056] Therefore, the above-mentioned frequency factor, activation energy, gas constant, and absolute temperature are extracted from a preset scheme by identifying the type of the current test chip and the corresponding load to determine the degradation rate of the current chip under the corresponding test.
[0057] When the corresponding function of the chip degrades, activation energy can be understood as the energy required for the physical or chemical process that causes the performance degradation. For example, the performance of some materials degrades faster at high temperatures, which may be because high temperatures provide enough activation energy, making it easier for certain structures or chemical bonds inside the material to change, thus causing performance degradation.
[0058] The scanning time of the chip and the corresponding number of channel connections are obtained in step S4 in order to determine the factors affecting the test efficiency and the test time under the corresponding time.
[0059] The length of the scan time directly affects the efficiency of the test; a shorter scan time may lead to incomplete testing and thus miss some potential problems; while too long a scan time can improve the accuracy of the test, it will reduce the test efficiency and increase the test cost; therefore, reasonably determining the scan time is the key to improving the test efficiency while ensuring the test accuracy.
[0060] The channel connection volume refers to the number of connections between various parts or modules in the chip; in chip testing, it is necessary to ensure that these channels can work properly, so each channel needs to be tested. The number of channel connections will directly affect the workload and complexity of the test; the larger the number of connections, the more channels need to be tested, and the test time will increase accordingly.
[0061] Through reasonable scan time and comprehensive testing of channel connections, the fault point in the chip can be located more accurately; for example, in boundary scan technology, by loading appropriate test patterns and observing the propagation path of the fault signal on the scan chain, the location of the fault can be accurately located, thereby accelerating the troubleshooting process.
[0062] like Figure 3 As shown, the processing method for identifying the problem area inside the chip in step S5 includes: S51, obtaining the error running instance tracking and the failed running instance tracking corresponding to the time length of the processing task, and obtaining the instance tracking results corresponding to the error running and the failed running; the instance tracking results obtained include the category labels and test parameters corresponding to the error running and the failed running.
[0063] S52, for the acquired instance tracking result, performing path resolution on the functional area corresponding to the instance tracking result, and obtaining a plurality of expected output path resolution results.
[0064] The path parsing method is to associate the obtained instance tracking results with the channel connection quantity and the channel information corresponding to the channel connection quantity to obtain the expected output path, mark the expected output path according to the category labels of erroneous operation and failed operation, and use the marked expected output path as the output expected output path parsing result.
[0065] S53, associating the instance tracking result with the degradation coefficient to obtain an expected degradation result related to the degradation coefficient.
[0066] The expected degradation result is to map the corresponding test parameters in the instance tracking result with the degradation coefficient to obtain the test parameters related to different degradation coefficients.
[0067] S54, repairing each functional area according to the expected degradation result of each functional area and the expected output path analysis result, and determining the repair instance result of the functional area.
[0068] Error running instance tracking and failed running instance tracking are performed through the running data of error codes or error prompts or failure prompts in the instance running process. When error running and failed running are carried out at the same time, error running instance tracking refers to nodes that have not experienced failed running, which means that data tracking has failed, or nodes that cause identification errors. Nodes with failed running are classified as failed running instance tracking.
[0069] The implementation method of repairing the instance result in step S54 includes: extracting the category labels of the erroneous operation and failed operation with the highest indicator value in the instance tracking result, taking the category label with the highest indicator value in the category labels of the erroneous operation and failed operation as the initial label, determining the occurrence probability of the initial label and the variation range of the initial label indicator under different instance tracking results, taking the initial label with the highest occurrence probability as the unique tracking label of the corresponding instance tracking result, and based on the variation range of the unique tracking label indicator, screening out the repair plan corresponding to the unique indicator from the preset repair plans, and repairing the instance tracking result.
[0070] When repairing the instance tracking results, the instance tracking results are repaired according to the currently selected repair plan and the execution order corresponding to the repair plan, and the adjustment range of the test parameters repaired in the instance tracking results is determined. According to the execution order of the instance tracking result repair and the adjustment range of the repaired test parameters, it is judged whether the current repair plan is a static cycle repair or a dynamic cycle repair. When it is a static cycle repair, the repair content of the previous repair plan is used as the repair content of the current instance repair result. When it is a dynamic cycle repair, according to the execution order of the instance tracking result repair and the adjustment range of the repaired test parameters, the average value of the adjustment range of the test parameters in the adjacent execution order is selected as the repair content of the current instance tracking result.
[0071] Static loop repair means that during the repair process, the instance tracking results are repaired according to a predetermined, fixed order and strategy, and the parameter adjustment range of each repair is also predetermined without significant changes.
[0072] Dynamic loop repair means that during the repair process, the execution order of the repair steps and the parameter adjustment range may be dynamically adjusted according to the actual situation in the repair process.
[0073] In step S6, the dynamic processing volume and performance degradation of the port are monitored in real time to adjust the chip test strategy.
[0074] The dynamic processing capacity represents the change in the actual processing capacity under different circumstances, and the performance degradation represents the change in the degradation coefficient corresponding to the dynamic processing capacity under specific conditions, so as to determine the adjustment range of the dynamic processing capacity when the current chip is degraded.
[0075] like Figure 4 As shown, the specific processing method is as follows: S61, generate a performance change curve according to the dynamic processing volume and performance degradation of the port; the horizontal axis of the performance change curve represents time, and the vertical axis represents the actual processing volume and the test parameters corresponding to the chip. The test parameters here are mainly used to mark the test parameters used by this processing method when the actual processing volume undergoes corresponding performance degradation; the actual processing volume will be marked on the performance change curve according to the different degradation coefficient ranges to identify the current performance degradation situation. At this time, the actual processing volume will include the performance degradation situation and the corresponding test parameter values.
[0076] S62, obtain the extreme points in the performance change curve, take the time point corresponding to the extreme point of the performance degradation that first appears as the first time point, take the time point corresponding to the extreme point closest to the first time point after the performance degradation occurs as the second time point, and obtain the local slope of the curve slope between the first time point and the second time point and the average slope of the corresponding data values.
[0077] Between the first time point and the second time point, multiple equally divided points are selected, and for each equally divided point, the slope of the line connecting the equally divided point and the adjacent points before and after is calculated, and the calculated slope of the line is used as the local slope. This will give the change in the local slope of the change value of the slope of the curve between the first time point and the second time point, so as to capture the reduction ratio of the corresponding parameter when performance degradation occurs, compare these local slope values, and determine the maximum or minimum value of the local slope value, which represents the maximum change in the performance degradation rate.
[0078] The average slope is determined by comparing the actual treatment volume or change in the test parameter value between a first time point and a second time point, which can be accomplished by calculating the difference between the two points and dividing by the time difference.
[0079] S63, comparing whether the maximum value of the local slope is within a preset range. If it is within the preset range, adjusting the chip test strategy based on the average value of the local slope. If it is not within the preset range, adjusting the chip test strategy based on the obtained average slope.
[0080] The processing method for adjusting the chip test strategy in step S63 includes: obtaining preset parameters corresponding to the local slope average value, using the preset parameters to interpolate the test parameters in the performance change curve, and outputting the interpolated test parameters as adjusted data; when the performance change curve is adjusted using the average slope, obtaining the preset parameter weight corresponding to the average slope, using the preset parameter weight to weighted average the test parameters in the current performance curve, and outputting the weighted average test parameters as adjusted data.
[0081] When setting the chip adjustment strategy, it also includes adjusting the test frequency. When adjusting the test frequency, an integral item corresponding to the performance change curve is added. Based on the error value of the performance change curve, the integral item output is determined. Based on the integral item output, the test frequency is adjusted.
[0082] The formula for the integral term is as follows: ;in, is the output of the integral term, is the integral gain, which is an adjustment parameter used to control the response speed of the integral term to the error. Is the error The integral over time t is the cumulative error of the performance change curve from the initial time to the current time; the error It indicates the error between the value of the corresponding performance change curve and the target value at time t.
[0083] In practical applications, since the thermal sensitivity coefficient works in a discrete time system (for example, performance data is read every certain period of time), the thermal sensitivity coefficient needs to use a discrete integral formula. The discrete integral can be approximated as the cumulative sum of errors, so the integral output is expressed as: ;in, is the integral output at the current time n, is the integral output of the previous moment n-1, is the integral gain, is the error between the value of the corresponding performance change curve and the target value at the current time n, is the time interval between two consecutive moments.
[0084] This formula means that at each time step Δt, the current error e[n] is multiplied by the integral gain , and add it to the integral output of the previous moment On the top, get the integral output at the current moment , so that the integral term can gradually adjust the output to eliminate the steady-state error of the system.
[0085] The output of the integral term is a control signal that is calculated based on the integral of the error over time. When the system has a steady-state error (i.e., a persistent deviation between actual performance and target performance), the integral term continues to accumulate the error and produces a gradually increasing control signal. This control signal is used to adjust the test frequency to gradually reduce the steady-state error.
[0086] Examples of adjusting test strategies: Example 1: If the performance of a problem area degrades rapidly and exceeds the preset threshold, you can consider increasing the test frequency of this area. For example, you can adjust the original test once a day to once an hour, so as to promptly discover and resolve potential problems.
[0087] Example 2: If the performance degradation of a problem area is related to specific test parameters or workloads, you can adjust these parameters or load conditions to observe performance changes, for example, reduce the operating voltage or frequency to see if performance improves.
[0088] Example 3: If a problem area involves multiple functional modules or ports, you can consider adjusting the test strategy for the entire module or port group, for example, adding test items to the module or adopting stricter test standards to ensure that the performance of the entire module is stable and reliable.
[0089] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations on the present invention. A person skilled in the art may make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention and they are still covered by the protection scope of the present invention.
Claims
1. A chip testing and analysis method, characterized in that: include: S1, determining multiple functional areas of the chip to be tested after packaging, and testing the input and output ports of the chip to be tested according to a preset allocation rule; S2, determining the actual processing capacity, processing interface information and test parameters used by the chip when testing the input and output ports; S3, determining the degradation coefficient of the chip under different workloads according to the actual processing amount and processing interface information of the chip; S4, setting the scanning time of the chip, and recording the number of channels that the chip can connect to within the scanning time as the channel connection quantity; determining the test signal and test protocol corresponding to each chip; S5, identifying the problem areas inside the chip based on the length of time the chip spends on processing tasks, the degradation coefficient, and the amount of channel connections; The processing method for identifying the problem area inside the chip in step S5 includes: S51, obtaining error running instance tracking and failed running instance tracking corresponding to the time length of the processing task, and obtaining instance tracking results corresponding to the error running and failed running; the obtained instance tracking results include category labels and test parameters corresponding to the error running and failed running; S52, performing path parsing on the functional area corresponding to the instance tracking result, with respect to the obtained instance tracking result, to obtain multiple expected output path parsing results; The path parsing method is to associate the obtained instance tracking results with the channel connection quantity and the channel information corresponding to the channel connection quantity to obtain the expected output path, mark the expected output path according to the category labels of error operation and failed operation, and use the marked expected output path as the output expected output path parsing result; S53, associating the instance tracking result with the degradation coefficient to obtain an expected degradation result related to the degradation coefficient; S54, repairing each functional area according to the expected degradation result of each functional area and the expected output path analysis result, and determining the repair instance result of the functional area; S6, monitor the dynamic processing volume and performance degradation of the ports in the problem area in real time and adjust the chip test strategy; The specific processing method of step S6 is as follows: S61, generating a performance change curve according to the dynamic processing volume and performance degradation of the port; S62, obtaining extreme value points in the performance change curve, taking the time point corresponding to the extreme value point of the first performance degradation as the first time point, taking the time point corresponding to the extreme value point closest to the first time point after the performance degradation occurs as the second time point, and obtaining the local slope of the curve slope between the first time point and the second time point and the average slope of the corresponding data values; S63, comparing whether the maximum value of the local slope is within a preset range. If it is within the preset range, adjusting the chip test strategy based on the average value of the local slope. If it is not within the preset range, adjusting the chip test strategy based on the obtained average slope.
2. A chip testing and analysis method according to claim 1, characterized in that: In step S2, when obtaining the actual processing capacity and processing interface information of the chip, the processing method for the chip includes: S21, setting test parameters of input and output ports according to chip design specifications and application scenarios, including data transmission rate, interface type, and signal level; S22, sending a preset signal sequence to the input port of the chip through the test software; S23, collecting and recording the actual processing volume of the chip in real time; comparing the collected actual processing volume with the preset processing volume of the chip to verify whether the actual processing volume corresponding to the corresponding processing interface information meets expectations.
3. A chip testing and analysis method according to claim 2, characterized in that: In step S23, the processing method for verifying whether the actual processing volume corresponding to the corresponding processing interface information meets the expected processing method includes: A preset signal sequence is sent based on the input port of the chip, and a first test result about the signal sequence is obtained. The first test result includes a basic processing volume that the chip can handle under the current workload, and based on the type and protocol of the corresponding processing interface information, a ratio of the basic processing volume of the first test result to the actual processing volume is determined. When the ratio of the basic processing volume of the first test result to the actual processing volume is less than the preset processing ratio, the difference between the actual processing volume and the preset processing volume under different workloads is determined, and a difference change curve is generated; based on the change trend of the difference change curve over time, the expected effect corresponding to the actual processing volume is determined; if the ratio of the basic processing volume of the first test result to the actual processing volume is greater than the preset processing ratio, the corresponding processing interface information and the corresponding input and output ports are identified, and the corresponding processing interface information and input and output ports are used as output.
4. A chip testing and analysis method according to claim 1, characterized in that: The degradation coefficient obtained in step S3 includes a performance degradation coefficient and a thermal sensitivity performance coefficient; The performance degradation coefficient is obtained by obtaining the degraded current, power and delay under the corresponding workload, and integrating the index values of the degraded current, power and delay into a comprehensive degradation coefficient as the performance degradation coefficient; The performance degradation coefficient is expressed as: ; in, is the performance degradation coefficient, is the initial power, is the degraded power, is the power adjustment factor, is the initial current, is the degraded current, is the regulation factor of the current, is the initial delay, is the degraded delay, is the adjustment factor for the delay.
5. A chip testing and analysis method according to claim 4, characterized in that: A thermal sensitivity performance coefficient, wherein the thermal sensitivity performance coefficient is determined according to the temperature obtained under the corresponding workload; and a degradation rate corresponding to the thermal sensitivity performance coefficient is determined based on a rate of change of the thermal sensitivity performance coefficient; ; in, is the thermal sensitivity coefficient, is the initial operating temperature, is the ambient temperature, is the rate of change of temperature with time; The degradation rate for acquisition is expressed as: ; in, is the rate of change of thermal sensitivity coefficient with temperature, is the degradation rate, is the frequency factor, is the activation energy, is the gas constant, is the absolute temperature.
6. A chip testing and analysis method according to claim 1, characterized in that: The implementation method of repairing the instance result in step S54 includes: Extract the category labels of the erroneous runs and failed runs with the highest indicator values in the instance tracking results, take the category label with the highest indicator value among the category labels of the erroneous runs and failed runs as the initial label, determine the occurrence probability of the initial label and the variation range of the initial label indicator under different instance tracking results, take the initial label with the highest occurrence probability as the unique tracking label of the corresponding instance tracking result, and based on the variation range of the unique tracking label indicator, screen out the repair plan corresponding to the unique indicator from the preset repair plans, and repair the instance tracking result.
7. A chip testing and analysis method according to claim 1, characterized in that: The processing method of adjusting the chip test strategy in step S63 includes: Obtain the preset parameters corresponding to the local slope average value, use the preset parameters to interpolate the test parameters in the performance change curve, and output the interpolated test parameters as the adjusted data; when the performance change curve is adjusted using the average slope, obtain the preset parameter weights corresponding to the average slope, use the preset parameter weights to weighted average the test parameters in the current performance curve, and output the weighted average test parameters as the adjusted data.
8. A chip testing and analysis method according to claim 1, characterized in that: When setting the chip adjustment strategy, it also includes adjusting the test frequency. When adjusting the test frequency, an integral item corresponding to the performance change curve is added. Based on the error value of the performance change curve, the integral item output is determined. Based on the integral item output, the test frequency is adjusted. The integral term output is expressed as: ; in, is the integral output at the current moment n, is the integral output of the previous moment n-1, is the integral gain, is the error between the value of the corresponding performance change curve and the target value at the current time n, is the time interval between two consecutive moments.
Citation Information
Patent Citations
Chip testing methods, apparatus, equipment and storage media
CN115629296B
Chip testing method and device and electronic equipment
CN116774018A
Method for monitoring aging process of working state of optical module in real time
CN109000734A
Pretreatment equipment for semiconductor packaging and use method
CN114749397A
Chip test method, test system, processor and memory medium
CN117316253A
